11 year old 5d MK3 superior to newest releases

[TomTheDP]

I have been thinking about this a lot. Why can't current cameras do 14 bit 4k raw?

The 5d MK3 featured a processor capable of 14 bit RAW at 3.5k, that was 11 years ago. I guess color depth isn't marketable? Although 10 bit seemed to be and still seems to be a huge selling point.

All these years later and no cameras have this feature outside of cinema cameras which have been doing this since 2010. To me that is what is missing in the lower end cameras. I have been tempted to get an old 5d MK3 and run magic lantern on it as a B-cam to my Alexa. Obviously there is a dynamic range difference but the color information is there. The workflow and ergonomics simply aren't good enough for me to use the camera on professional projects.

 

[PannySVHS]

Always good to see what a 5d can do in the hands of a Dop. 🙂

[PPNS]

its pixel binned since that sensor isnt natively 3.5k, so you should actually be missing quite a lot of info right?

[TomTheDP]

3 hours ago, PPNS said:

its pixel binned since that sensor isnt natively 3.5k, so you should actually be missing quite a lot of info right?

My Sigma FP does a similar thing not sure if its pixel binning or line skipping or if there is a difference between the two. It is 4k rather than 3.5k but not far off. Similar camera, 11 years newer, but only 12 bit readout.

The Sigma FP post workflow isn't the greatest but at least CDNG can be immediately put into resolve. But yeah not 14 bit and definitely lacking when comparing to a RED or Alexa, including a 10 year old RED or Alexa.

@mercerowns both the FP and 5D MK3 and I believe he has said the lower bit depth on the Sigma is noticeable. I have heard that from others too.

[FHDcrew]

Also the 5D III does what many of us have wanted from modern cameras but haven’t gotten because of dumb politics…

 

Uncompressed RAW sensor data dumped to the card. Nothing fancy. No compression, just taking the data and dumping it. Think of how many cameras have been released in the past 5 years with fast CFExpress interfaces. Could easily write RAW video to the card. This is so simple, even the 5d Mark II and the ancient Canon 50D do this. In 14 bit too. Not even hard. Instead they make us send the feed via HDMI to an external recorder when we are using modern cameras. 

[billdoubleu]

I suspect it's more a matter of why would a large Japanese manufacturer provide such a feature. There doesn't seem to be much incentive to include high bit depth RAW in the compact mirrorless market. It seems people are demanding that bodies get smaller, more lightweight, and provide higher resolution (I know the last one is debatable). Smaller bodies introduce heat management and battery life issues. Higher resolution introduces more data to process and store. Miniaturization costs money.

The manufacturer also has to consider the goal of the end user. The people on this forum are regular guys and gals busting their asses day to day competing in a thankless market where there is always someone who will do the job cheaper and faster. I think most want to get as much done in-camera as possible and get it out the door. Not every project should involve processing and coloring footage. Most one man bands or small production teams just don't have the time or resources for that.

In one of Andrew's latest front page articles he is proving that codecs are getting really good and flexible and the data rates are far more manageable than RAW. Ten years ago codecs were either shit at low data rates or on the flip side you needed a beefy computer to handle the high data rate codecs.

From what I understand, and I could be wrong, the majority of productions using a cinema camera like an ARRI don't use a RAW workflow, aside from scenes with a challenging lighting situation. The data collection and management is too much for all but the biggest budgets. And any niche cases are being handled by the likes of Blackmagic with their sort-of-RAW or that other company we all know who owns the compressed RAW market (which in and of itself is enough reason for manufacturers to not want to deal with it).

So basically I don't think enough people give a shit about 14 but RAW for it to be a readily available option. If there is one thing I took away from my economics courses is that incentive is everything.

[TomTheDP]

39 minutes ago, billdoubleu said:

I suspect it's more a matter of why would a large Japanese manufacturer provide such a feature. There doesn't seem to be much incentive to include high bit depth RAW in the compact mirrorless market. It seems people are demanding that bodies get smaller, more lightweight, and provide higher resolution (I know the last one is debatable). Smaller bodies introduce heat management and battery life issues. Higher resolution introduces more data to process and store. Miniaturization costs money.

The manufacturer also has to consider the goal of the end user. The people on this forum are regular guys and gals busting their asses day to day competing in a thankless market where there is always someone who will do the job cheaper and faster. I think most want to get as much done in-camera as possible and get it out the door. Not every project should involve processing and coloring footage. Most one man bands or small production teams just don't have the time or resources for that.

In one of Andrew's latest front page articles he is proving that codecs are getting really good and flexible and the data rates are far more manageable than RAW. Ten years ago codecs were either shit at low data rates or on the flip side you needed a beefy computer to handle the high data rate codecs.

From what I understand, and I could be wrong, the majority of productions using a cinema camera like an ARRI don't use a RAW workflow, aside from scenes with a challenging lighting situation. The data collection and management is too much for all but the biggest budgets. And any niche cases are being handled by the likes of Blackmagic with their sort-of-RAW or that other company we all know who owns the compressed RAW market (which in and of itself is enough reason for manufacturers to not want to deal with it).

So basically I don't think enough people give a shit about 14 but RAW for it to be a readily available option. If there is one thing I took away from my economics courses is that incentive is everything.

I agree most things in life come down to money. I do question the why though.

The Alexa processes in 14 bit for prores as well, not really any difference between Prores XQ and RAW. RAW is a gimmick to me. The whole benefit of REDraw is not that its RAW but that it is capturing 16 bits of color depth. Throw any 10 bit log with a decent data rate thru the correct image pipeline and it will look just as good as the 12 bit raw from these cameras because the readout is no different. Sometimes the RAW looks worse due to less processing and more color noise.

The RAW gimmick has worked for RED for the past 12 years, who thrive in the owner operator market. Same with Blackmagic which have always touted their RAW recording capabilities, even if BRAW is really just 12 bit log. The new RED Raptor having 444 XQ is actually a huge selling point for me, as to me RAW is just a hassle.

All the companies are pushing for RAW. Apple prores RAW, Nikon even if putting RAW video in their cameras as a selling point and apparently it's worth the hassle of lawsuits with RED. Almost all cameras now record RAW to external recorders if not internal. It is a huge part of the entire Atomos brand, RAW recorders.

10 bit log recording is now a standard in pretty much every camera even Go Pros, even a lot of phones have it now. In a world where cell phones are the go to device, I think every feature is important as you need something to make it worth while lugging a camera around that doesn't fit in your pocket.

Anyways my thought is that there is a push for better IQ on all fronts; resolution but also, dynamic range, and color depth. However no one has caught up to an 11 year old hacked Canon DSLR. To me bit depth is the major thing lacking IQ wise in non cinema cameras. Dynamic range is about good enough as is resolution. Give me 14 bit rather it be RAW or prores or whatever codec you want to package it in. Hopefully not CDNG but hey I'll take that too.
 

[Attila Bakos]

10 hours ago, PPNS said:

its pixel binned since that sensor isnt natively 3.5k, so you should actually be missing quite a lot of info right?

The maximum I could get out from my 5D3 is 3584 x 1730 in 14bit lossless. That's not binned, it uses a 1.6x crop.
We are now at around 150MB/s maximum with card spanning, sd card overclocking and several hacks to boost performance. And some talented people are still tweaking this thing.

[OleB]

For the fp I can answer this. The Sony IMX410 sensor is not capable of outputting 14bit in a high enough frame rate. 19 fps is roughly the maximum. Additionally that sensor only has a 14 bit readout mode for the full sensor. No pipeline for less pixels and faster 14bit … 

[Ty Harper]

While I have moved on to the R5C (and either an R5 and R6 MKII as a b cam in the future!) - I still have two 5D MKIII's and don't think I'll ever part with them. Look the look, and the workflow is so much easier in the post M1 era!

[PPNS]

- i think its hard for me to believe that the software hack gives you superior image quality to anything now

- its hard for me to believe that the 14 bit readout would give you any discernible difference compared to the 12 bit readout

[TomTheDP]

27 minutes ago, PPNS said:

- i think its hard for me to believe that the software hack gives you superior image quality to anything now

- its hard for me to believe that the 14 bit readout would give you any discernible difference compared to the 12 bit readout

It gives you 14 bit color which is a lot more color information than 12 bit. It just depends on what part of IQ is important to you. Past 2k isn't important to me, dynamic range is important, but color is the most important. Dynamic range on cameras like the Lumix S1 have been good enough for me. About a stop and a half behind an Alexa but good enough. Another example, the Lumix S1 can do 12 bit raw/10 bit log and has more dynamic range/higher resolution than the 10 year old RED scarlet. However I found the color overall out of the old Scarlet looked much richer with nicer skintones. The colorist on that project told me that as well. I am sure this is partially color science. The Scarlet sensor is quite old though and was never known to have the best color science. I do attribute a lot of it to 16 bit depth.

Its not as much side by side comparisons but having shot and colored features on low end cameras like Blackmagic, Z-cam, Panasonic, Fuji and than doing the same with ARRI or RED the difference is very noticeable. Obviously this isn't a scientific observation.

RED and ARRI are high end cinema cameras. So it may just be that they have much more pleasing color science and that's it. But I have also observed still images coming out of these hybrid cameras that always look superior to the video they capture. 14 bit vs 12-10 bit readout.

I'd have to find the thread but it was a gent on here who shot some material with the BM 4K, C70, and his old 5D MK3. He was expecting the newer cameras to replace his old Canon but was left quite disappointed.

 

[billdoubleu]

23 hours ago, TomTheDP said:

I agree most things in life come down to money. I do question the why though.

The Alexa processes in 14 bit for prores as well, not really any difference between Prores XQ and RAW. RAW is a gimmick to me. The whole benefit of REDraw is not that its RAW but that it is capturing 16 bits of color depth. Throw any 10 bit log with a decent data rate thru the correct image pipeline and it will look just as good as the 12 bit raw from these cameras because the readout is no different. Sometimes the RAW looks worse due to less processing and more color noise.

The RAW gimmick has worked for RED for the past 12 years, who thrive in the owner operator market. Same with Blackmagic which have always touted their RAW recording capabilities, even if BRAW is really just 12 bit log. The new RED Raptor having 444 XQ is actually a huge selling point for me, as to me RAW is just a hassle.

All the companies are pushing for RAW. Apple prores RAW, Nikon even if putting RAW video in their cameras as a selling point and apparently it's worth the hassle of lawsuits with RED. Almost all cameras now record RAW to external recorders if not internal. It is a huge part of the entire Atomos brand, RAW recorders.

10 bit log recording is now a standard in pretty much every camera even Go Pros, even a lot of phones have it now. In a world where cell phones are the go to device, I think every feature is important as you need something to make it worth while lugging a camera around that doesn't fit in your pocket.

Anyways my thought is that there is a push for better IQ on all fronts; resolution but also, dynamic range, and color depth. However no one has caught up to an 11 year old hacked Canon DSLR. To me bit depth is the major thing lacking IQ wise in non cinema cameras. Dynamic range is about good enough as is resolution. Give me 14 bit rather it be RAW or prores or whatever codec you want to package it in. Hopefully not CDNG but hey I'll take that too.
 

Perhaps it mostly still comes down to cost and reliability? Mainly in regards to storage media. Cameras are only very recently able to take advantage of low(ish) cost SSDs that are fast and of large capacity. In the recent past storage media was slow, low capacity, and expensive; or slow, high capacity, and really expensive. Couple that with the variance in card speed and card compatibility with different cameras, even with the same card brand. It's difficult to ensure a quality customer experience with these issues.

Also, you're an experienced DP, you may be more willing to work with difficult equipment (like a hacked Canon) to aquire a better image. I would venture to say that the majority of these cameras end up in the hands of jackoffs like me who participate in this as a hobby and don't have the inclination to dick around with a finicky camera.

RED caught a lot of shit for charging astronomical prices for their branded SSDs but it makes sense if you are quality assuring reliability and compatibility. I'm not sure how ARRI handles this aspect of their cameras but I'm sure it's much of the same. Limit the customer to what you're sure will work even though the cost is high. In these two cases you can do that, but with a consumer camera you will probably account for the lowest common denominator and accommodate accordingly. Democracy at it's finest.

[TomTheDP]

12 hours ago, billdoubleu said:

Perhaps it mostly still comes down to cost and reliability? Mainly in regards to storage media. Cameras are only very recently able to take advantage of low(ish) cost SSDs that are fast and of large capacity. In the recent past storage media was slow, low capacity, and expensive; or slow, high capacity, and really expensive. Couple that with the variance in card speed and card compatibility with different cameras, even with the same card brand. It's difficult to ensure a quality customer experience with these issues.

Also, you're an experienced DP, you may be more willing to work with difficult equipment (like a hacked Canon) to aquire a better image. I would venture to say that the majority of these cameras end up in the hands of jackoffs like me who participate in this as a hobby and don't have the inclination to dick around with a finicky camera.

RED caught a lot of shit for charging astronomical prices for their branded SSDs but it makes sense if you are quality assuring reliability and compatibility. I'm not sure how ARRI handles this aspect of their cameras but I'm sure it's much of the same. Limit the customer to what you're sure will work even though the cost is high. In these two cases you can do that, but with a consumer camera you will probably account for the lowest common denominator and accommodate accordingly. Democracy at it's finest.

That is true. The sigma FP is limited by write speed to some degree.

I am willing to deal with something finicky, but it is project dependent. I appreciate the simplicity of Prores over anything else as it makes post very very simple. I am considering using the Sigma FP for an upcoming feature, but I know I will have complete control over the post production process.

I think RED caught a lot of flack because after that one individual opened up their SSD's they were found to be SSD's not specially made for RED. I think the mark up for there SSD's was like 8x what it would otherwise cost.

I personally don't mind proprietary products. For example, the Alexa mini can only use ARRI branded CFAST cards. The LF and XT only use specific Alexa SSD's, which are very costly. From what I have seen RED has just not been the most honest and I really think if they were more truthful they'd actually have a better reputation.
 

[Andrew Reid]

On 3/16/2023 at 10:01 PM, PPNS said:

its pixel binned since that sensor isnt natively 3.5k, so you should actually be missing quite a lot of info right?

The 5D III doesn't pixel bin in Magic Lantern 3.5K RAW

It is a S35 crop of the sensor (1:1 readout) with no binning or line-skipping.

And yes, it's bloody amazing.

[zerocool22]

Yeah this is a shootout that I want to see, I do miss my 5D III. If I can find one for a bargain I might pick it up and compare it to the S5 II. The micro jitter of the 5D III is something that I do not miss though, but locked off it might be interesting. 

[mercer]

On 3/16/2023 at 9:57 PM, TomTheDP said:

mercerowns both the FP and 5D MK3 and I believe he has said the lower bit depth on the Sigma is noticeable. I have heard that from others too.

Not only can I notice the lower bit depth, I'm fairly certain there is some type of noise reduction occurring behind the scenes with the FP. The extra resolution from the FP is nice but I feel like it resembles a lot of other camera's images from compressed codecs with just more leeway in post to make some changes.

The 5D3, on the other hand, has a higher end look in my opinion and the image is more malleable especially in the highlights, but the shadow detail is much cleaner with the FP.

They also share a similar post workflow for me... bring the footage into Resolve, correct/grade, export as ProRes and import into FCPX for the final edit/grade.

At first glance, the bump in resolution may make you feel like the FP has a superior image, but once you get into the files... let's just say that if I had to choose one, I'd sell the FP and keep the 5D3.

With that said, I think people underestimate the benefits of 14bit color. In some ways, it gives a perceived bump in DR because it is capturing the minute differences in hues. I notice it in skies and in tree bark.

I'm also a fan of shooting with DSLRs... especially chunky, full frame ones. If I could afford a 1DX iii, I'm sure I would love it.

[PannySVHS]

Hehe, couldnt resist and had to get one. A 250.000 shutter though, but still working. For 300EU. Got a 5D2 for much cheaper two months ago. In great shape, 30.000 shutter, for 150. Seller contacted me to give me a nifty fifty 1.8 and a nice backpack for free. Anyway, now I gotta use these beauties. Any quick and tasty infos on the differences? Some prefer the look from the 5D2? 5D3 much more hasslefree to use in Magic Lantern RAW due to dual card slots? cheers

[rosco]

You’re using the Sigma fp as a B cam anyway, I don’t think audience would care if it’s missing the 2 extra bits in raw. I think when the lighting is good and the color closely matches, non pixel peeping audiences wouldn’t really notice.  I would only use the B cam for certain angles and pick up shots, A cam would be used 90-95%.  Ymmv.

[TomTheDP]

2 minutes ago, rosco said:

You’re using the Sigma fp as a B cam anyway, I don’t think audience would care if it’s missing the 2 extra bits in raw. I think when the lighting is good and the color closely matches, non pixel peeping audiences wouldn’t really notice.  I would only use the B cam for certain angles and pick up shots, A cam would be used 90-95%.  Ymmv.

I mean I probably wouldn't shoot with a Classic if my only concern was audience perception.

 

3 hours ago, mercer said:

Not only can I notice the lower bit depth, I'm fairly certain there is some type of noise reduction occurring behind the scenes with the FP. The extra resolution from the FP is nice but I feel like it resembles a lot of other camera's images from compressed codecs with just more leeway in post to make some changes.

The 5D3, on the other hand, has a higher end look in my opinion and the image is more malleable especially in the highlights, but the shadow detail is much cleaner with the FP.

They also share a similar post workflow for me... bring the footage into Resolve, correct/grade, export as ProRes and import into FCPX for the final edit/grade.

At first glance, the bump in resolution may make you feel like the FP has a superior image, but once you get into the files... let's just say that if I had to choose one, I'd sell the FP and keep the 5D3.

With that said, I think people underestimate the benefits of 14bit color. In some ways, it gives a perceived bump in DR because it is capturing the minute differences in hues. I notice it in skies and in tree bark.

I'm also a fan of shooting with DSLRs... especially chunky, full frame ones. If I could afford a 1DX iii, I'm sure I would love it.

How is monitoring with the 5D MK3? Really considering it honestly.